NVIDIA’s Big Week, New AI Models, Social Media On Trial

Welcome back to Big Technology’s Agenda Setter email, in your inbox most Mondays with a preview of what’s to come in the week ahead.

Nvidia reports earnings this Wednesday and the stakes will be tremendous as usual. CEO and co-founder Jensen Huang is expected to give updates on inference demand, the pace of shipped compute, and more on the company’s vision for robotics.

In a rocky month for AI (and the software stocks it might destroy), Wall Street’s bar for Nvidia is especially high, with some expecting between 60% and 70% year-over-year revenue growth. Nvidia stock has barely moved since the start of last year’s fourth quarter, it’s up 1.7% vs. the S&P’s 3.3%, and it might take a legendary report to cause it to budge.

So even as the tech giants have committed to spending $600 billion on infrastructure in 2026, the company’s earnings results will be a must-watch as usual.

There’s also a new wrinkle for Nvidia this week. After months of reports that the company’s relationship with OpenAI has been fraying, we may indeed see it come through on plans to invest $30 billion in OpenAI, part of a $100 billion funding round that is reportedly nearing a close.

If the OpenAI round is not announced this week, Huang might still provide some clarity (and likely steadfast support) about how he feels about Sam Altman and Co. Those comments may well be the week’s big headline.

One thing worth your attention:

It’ll be worth watching how Huang talks about supply and demand. On its last earnings call, when an analyst asked him if there’s a “realistic path” for supply to catch up with demand over the next 12 to 18 months, he didn’t give a straightforward answer. More clarity this time would be nice.

We’re also less than a month away from Nvidia’s annual GTC conference, where Huang said the company will “surprise the world” with a new kind of chip. Any new information on that front could make waves.

Other events this week:

• Other companies scheduled to report earnings this week, with each likely to give updates on their AI roadmaps:

• February 24: Workday, GoDaddy, HP

• February 25: Salesforce, Snowflake

• February 26: Dell, Intuit, Coreweave

• The U.S. Senate Foreign Relations Committee on Tuesday will have a briefing to discuss “tech competition, critical minerals, and supply chain security.”

• The Senate Judiciary Committee will meet on Feb. 26 to discuss several proposed bipartisan bills related to protecting children online, including the COP Act, the ECCHO Act and the SAFE Act.

The Intelligence Report

• Perplexity said it’s moving away from ads, claiming ads hurt trust in AI right as rival ChatGPT begins tests to introduce ads.

• Sam Altman and Dario Amodei refused to hold hands while appearing on stage at the India AI Impact Summit.

• ChatGPT’s early brands reportedly testing ads include Best Buy, Expedia, Qualcomm and Enterprise according to AdThena.

• Meta plans to spend $65 million this election cycle to fund AI-friendly state candidates.

• Bernie Sanders and Ro Khanna spoke at a Stanford town hall about billionaires and the future of AI.

• Chinese tech companies Alibaba and Moonshot released new AI models, which both feature text, video and image generation.

• Google and Apple announced new music focus features, with Gemini adding a way to create 30-second music tracks and Apple Music adding a way to make AI-generated playlists.

• ByteDance’s new AI-video generator Seedance 2.0 reportedly sent Hollywood into a panic, with the Motion Picture Association sending a cease-and-desist letter.

Your Mac: Calmer, Faster, and Tended From The Inside. (sponsor)

Your Mac works best when its inner space feels cared for. CleanMyMac helps you tidy clutter, prune outgrown files, and guard against malicious “bugs,” so everything runs smoother.

With CleanMyMac, you’ll see what’s taking up room, what’s slowing you down, and what’s safe to remove. Think of it as a calm, competent caretaker for your system, whether it’s for a quick sweep or a full tune-up. Use code Big20 and run your Mac better today:

Learn More

Under The Hood: Model Updates

Gemini 3.1 Pro:

• Google announced Gemini 3.1 Pro, which will power new capabilities for consumer apps like Gemini and NotebookLM along with various enterprise products and platforms. One of the things Google touted was its performance in benchmarks, including more than double the reasoning performance compared to the previous Gemini model.

• In a blog post, Google gave a few examples of how Gemini 3.1 Pro could power various tasks that benefit from advanced reasoning, including code-based animations and more advanced web design,

• Early enterprise clients testing Gemini 3.1 Pro include JetBrains, Databricks, Cartwheel and Hostinger Horizons.

• Some of Google’s updates seem to be positioning Gemini to further powers agentic workflows via data synthesis, long context, multi-step tools, and agentic coding.

Claude Sonnet 4.6

• Anthropic debuted its new Claude Sonnet 4.6 model, which it says is reaching capabilities similar to its larger Opus 4.6 model, but at Sonnet’s lower pricing. It’s also introducing Sonnet 4.6 with 1 million token context window in beta, which Anthropic says is big enough to hold entire “codebases, lengthy contracts or dozens of research papers in a single request.”

• Sonnet 4.6’s capabilities include improving agentic tools and upgrades in computer use that let the agent operate software like a human would. It also has better long context reasoning and better capabilities for business applications like financial analysis and document analysis. Although Opus 4.6 still beats Sonnet 4.6 in various benchmarks, the gap seems to be closing.

• Some of the platforms Sonnet 4.6 will update include Claude Code and Claude Cowork. When Anthropic asked users about their model preferences, it found users preferred Sonnet 4.6 over Opus 4.6, thanks in part to Sonnet’s better context awareness, fewer hallucinations and more consistent work with coding tasks.

Share

Mark Zuckerberg goes to court in Los Angeles

Meta CEO Mark Zuckerberg took the stand last week for a rare appearance in a civil trial that could alter how social media companies defend themselves in lawsuits alleging harm from platform design features.

The jury trial in Los Angeles, now entering its third week, is the first in a series of “bellwether” cases to test whether features like infinite scroll and algorithmic feeds can be treated as addictive product designs instead of neutral tools for organizing content.

Zuckerberg’s appearance came almost exactly two years after his 2024 appearance in Congress, where he infamously offered an impromptu apology to the families of kids targeted by social media abuse. Earlier in the trial, Instagram chief Adam Mosseri also testified and answered questions about whether engagement features can meaningfully contribute to addiction-like behavior.

The plaintiff in this case is a 20-year-old woman identified as K.G.M., who is one of roughly 1,600 plaintiffs in numerous cases alleging that Meta and other companies engineered engagement-driven features leading to compulsive use by minors and contributed to anxiety and depression. YouTube is also a defendant in the lawsuits, but TikTok and Snap — which were both initially included as defendants — settled before the trial began.

The case hinges on whether social media companies’ engagement-driven features can be treated as product design choices instead of editorial decisions. Historically, the latter has been protected by Section 230 of the Communications Decency Act, which has largely allowed platforms to avoid being held liable for content on their platforms.

The case could also raise issues about generative AI could pose new risks as features become powered by large language models and other types of AI-generated content. Even beyond the trial, there’s growing concern about potential psychological harm from chatbots like ChatGPT and Claude, including “AI psychosis” or other types of emotional manipulation and misinformation. Of course, Meta, Google and others have already come under scrutiny for their own LLM-powered chatbots.

Just like the early days of social media, the impact of chatbots is still unclear, but experts are already trying to understand potential causes and correlations. Last month, Harvard researchers published findings from their 2025 survey of 20,000 U.S. adults that found daily or frequent use of AI was associated with higher levels of symptoms related to anxiety and depression. They also found frequent AI use was more common in some groups including men, younger adults, people in cities, and even people with higher levels of education and higher incomes.

The latest on Big Technology Podcast

• Box CEO Aaron Levie joined the podcast to talk about OpenAI’s rumored $100B raise, Anthropic’s new model, and OpenAI’s acquisition of OpenClaw.

  “I think you’re in a period right now where you’re just in the infrastructure build out, teach the world about AI. It’s sort of worth subsidizing a lot of these use cases because it’s the fastest path to figuring out where the actual value is going to be.”

  -Aaron Levie

• Wrapping up Big Technology’s Davos podcast series, Google DeepMind COO Lila Ibrahim and SVP James Manyika outlined Google’s evolving approach to R&D amid the intensifying AI race. Alongside AI’s role in education, they discussed products like NotebookLM and Workspace, as well as longer-term bets in quantum computing, materials science and even training AI in space.

You can listen on Apple Podcasts, Spotify, or your podcast app of choice

Join Big Technology’s Private Discord Server!

Where we’ll talk about this story, the latest in AI, the week’s podcast, and plenty more. You can sign up via the link below:...